A critical challenge facing emergency department (ED) physicians is how best to manage patients presenting with symptoms of heart failure. Currently, most patients being evaluated for heart failure are admitted to the hospital, yet not all of these patients warrant such intensive treatment, and up to 50% of these admissions could be avoided. Improving the ability of the emergency physician to effectively and safely manage low-risk patients is essential to avoid unnecessary hospitalizations. We propose developing a decision tool derived from prospectively gathered ED data that will predict risk for inpatient or outpatient death and serious in-hospital or out-of-hospital complications. Further, the proposed project will validate the usefulness and generalizability of this decision tool in three different ED environments across racially and socioeconomically diverse patient populations. To develop our decision tool, over 100 variables routinely available to the emergency physician within the first two hours of ED presentation will be considered for inclusion in a statistical risk model. Unlike exisitng models using inpatient data, these measures are representative of actual clinical practice and routinely used to decide a patient's disposition. We will collect standardized data during a patient's evaluation for heart failure. Relying on chart review or large dataset analyses can lead to missing and inconsistent data. We will include all patients evaluated for heart failure regardless of final diagnosis, thus avoiding selection bias inherent in models based on patients with a definitive diagnosis. A fundamental innovation we propose is a tool using 5-day outcomes for primary analyses, and 30-day outcomes for secondary analyses. This overcomes the limitation of 30-day outcome models that are highly dependent on unpredictable, post-visit patient and provider behavior. Another novel aspect of the proposed project is the combining of expertise in emergency medicine, cardiology, and biostatistics to accurately assign post-treatment outcomes to acute presentations. Results will be translated into an algorithm that will be disseminated worldwide. This is the first step toward achieving our broad objective of appropriate allocation of hospital resources to reduce costs of heart failure care. In collaboration with outcomes and effectiveness researchers, we plan to conduct further studies to test the efficacy of our risk model.